Type 2 Diabetes (T2D) is a complex disease with considerable inter-patient heterogeneity. T2D management could likely be improved through a precision medicine approach with identification of clinically distinct T2D subgroups. The goal of this NIH K23 research proposal is to determine whether available genetic and clinical data can be leveraged to define novel and reproducible T2D subtypes. While there are almost 100 known common T2D variants, far fewer rare T2D variants are known, and there is currently no known role for T2D genetic variants in patient management. Recent research has demonstrated that a continuum exists for phenotypes of patients with monogenic diabetes and T2D; therefore, understanding rare diabetes-related traits can likely inform upon the pathogenesis of T2D. This proposal asks whether novel and reproducible diabetes- related subtypes can be identified using i) genetics, with rare variants in Aim 1 and common variants in Aim 2, as well as by using ii) both rare and common genetic variation combined with clinical data in Aim 3. Aim 1 is to analyze exome sequences to identify causative mutations in families with rare insulin secretion abnormalities; we hypothesize that rare variants in these genes will also contribute to T2D risk, as evaluated in the largest available T2D exome study (55,000 exomes). Moving from rare to common genetics, Aim 2 is to apply Bayesian non-Negative Matrix Factorization (bNMF) to 88 known common T2D variants and associated traits from genome-wide association studies (GWAS) to identify shared biological pathways. These T2D genetic clusters will be tested for associations with diabetes-related outcomes, and then attempted to be translated into T2D patient subtypes using two large electronic health record (EHR)-linked Biobanks. Finally, Aim 3 is to take a patient-centered approach combining clinical and genetic data (both rare and common variants) to construct T2D subtypes, using bNMF clustering in large cohort studies and EHR-linked Biobanks. This research could potentially identify new mutations causing rare diabetes-related diseases and also new rare variants causing T2D, in addition to uncovering novel T2D subtypes. The proposed research ideally will help clarify the genotypic-phenotypic relationship of T2D genetic variation, offer a rational framework for application of genetic data into clinical care, and provide the training necessary for the Principal Investigator, Dr. Miriam Udler, to transition into research independence. Dr. Udler will apply her background in statistical genetics and attain new skills in exome sequence analysis in families, big-data clustering approaches, and diabetes physiology in order to ideally elucidate T2D pathophysiology and ultimately improve T2D management. Through her K23 mentored training, Dr. Udler intends to develop the skills necessary to devote her career to patient-oriented endocrine- genetics research.